Acidithiobacillus ferrooxidansis a gamma-proteobacterium that lives at pH2 and obtains energy by the oxidation of sulfur and iron. It is used in the biomining industry for the recovery of metals and is one of the causative agents of acid mine drainage. Effective tools for the study of its genetics and physiology are not in widespread use and, despite considerable effort, an understanding of its unusual physiology remains at a rudimentary level. Nearly complete genome sequences ofA. ferrooxidansare available from two public sources and we have exploited this information to reconstruct aspects of its sulfur metabolism.
Two candidate mechanisms for sulfate uptake from the environment were detected but both belong to large paralogous families of membrane transporters and their identification remains tentative. Prospective genes, pathways and regulatory mechanisms were identified that are likely to be involved in the assimilation of sulfate into cysteine and in the formation of Fe-S centers. Genes and regulatory networks were also uncovered that may link sulfur assimilation with nitrogen fixation, hydrogen utilization and sulfur reduction. Potential pathways were identified for sulfation of extracellular metabolites that may possibly be involved in cellular attachment to pyrite, sulfur and other solid substrates.
A bioinformatic analysis of the genome sequence ofA. ferrooxidanshas revealed candidate genes, metabolic process and control mechanisms potentially involved in aspects of sulfur metabolism. Metabolic modeling provides an important preliminary step in understanding the unusual physiology of this extremophile especially given the severe difficulties involved in its genetic manipulation and biochemical analysis.
Acidithiobacillus ferrooxidansis a chemolithotrophic, mesophilic, gamma-proteobacterium that lives at pH2. It is found in mine drainage and coal wastes and other acidic environments and is used extensively as part of a consortium of microorganisms for the industrial recovery of metals, such as copper and gold . It can obtain all its energy and electron requirements from the oxidation of various forms of reduced sulfur and ferrous iron. It can also anaerobically reduce sulfur using electrons derived from hydrogen and it can fix nitrogen and carbon dioxide.
The ability ofA. ferrooxidansto obtain energy and electrons from sulfur and reduced sulfur compounds has been investigated at the biochemical level. [2–5]. However, less attention has been paid to the mechanisms it uses to take up sulfur (sulfate) from its environment and how it is subsequently assimilated into organic compounds such as the amino acids methionine and cysteine, iron-sulfur centers in electron transfer proteins and various sulfated metabolites .A. ferrooxidansmust balance its requirements for sulfate for biosynthesis with its use of sulfur as an energy and electron source. Nothing has been described regarding the regulation of these requirements. Moreover, nothing is known about the possible coordinated regulation of sulfur metabolism and hydrogen utilization and nitrogen fixation that are known from studies in other organisms to impinge upon sulfur metabolism. Furthermore, little is known for any organism about mechanisms for sulfate uptake in acid environments, especially those inhabited byA. ferrooxidanswhere high concentrations of sulfate ions and potentially competing ions such as molybdenate are frequently found. Although the internal pH ofA. ferrooxidansis near neutral, its periplasm and outer membrane are exposed to pH2 and proteins in these locations must fold and function in high concentrations of protons. Questions also arise as to how membrane transport mechanisms function inA. ferrooxidanswhen confronted by a pH gradient that covers five logs of magnitude, especially considering that in neutrophilic organisms many of these transporters are symporters or antiporters using protons to drive uptake or discharge of metabolites.
Such questions could normally be addressed by a range of genetic and biochemical experiments. Unfortunately,A. ferrooxidanshas proved recalcitrant to standard genetic manipulation. There is only one report of transformation inA. ferrooxidansand this may be strain specific and not of general use  and transduction is unknown. Only recently have techniques for conjugation been established and these remain difficult to control and are of low efficiency . Exacerbating the problem is the difficulty of obtaining sufficient cell mass for many biochemical assays. Given these experimental hurdles, we argue that metabolic models derived from bioinformatic analyses offer an especially attractive starting point for unraveling the interesting physiology ofA. ferrooxidans. Such studies have already revealed valuable information regarding its amino acid biosynthesis , nitrogen metabolism, metal fluxes and other characteristics .
We have carried out a bioinformatic analysis of the genome sequence of the type strain ofA. ferrooxidansATCC23270 made available by The Institute for Genome Research and Integrated Genomics in an attempt to identify genes and proteins potentially involved in sulfur uptake and assimilation. Using this information, a preliminary model for sulfur assimilation inA. ferrooxidansis presented and possible mechanisms for its biochemical and genetic regulation are discussed. The model serves as a first attempt to understand global sulfur uptake and assimilation inA. ferrooxidans.
Results and Discussion
The results are divided into six sections for convenience of discussion. Section 1 presents the model and lists the genes that constitute the central part of the model. Section 2 describes two possible mechanism for sulfate uptake from the environment. Section 3 describes a potential pathway strongly predicted to be involved in the activation of sulfur and its incorporation into cysteine via adenosine-5'-phosphosulfate (APS). Suggestions are also made as to how this pathway might be regulated. Section 4 identifies a potential pathway for the sulfation of metabolites that might be involved in extra-cellular matrix formation via the PAPS (3'-phosphoadenosine 5'-phosphosulfate) pathway. In Section 5 a proposed connection between cysteine and the formation of Fe-S centers is discussed. We also speculate about the possible role of putative regulatory genes, embedded in predicted regulons for nitrogen fixation and hydrogen oxidation, in sulfur assimilation and sulfur reduction. In Section 6 we have tried to draw together the various postulated regulatory mechanisms in order to paint a more global picture of the regulation of sulfur metabolism inA. ferrooxidans, highlighting aspects of its chemolithoautotrophic physiology.
SECTION 1: Proposed preliminary model for sulfur uptake and assimilation
Figure1presents a preliminary model for sulfur uptake and assimilation inA. ferrooxidansderived from a bioinformatic analysis of the potential genes present in its genome. A list of the candidate genes presented in the model, their proposed functions and some of their characteristics are shown in Table 1.
SECTION 2: Possible sulfate uptake mechanisms
Microorganisms obtain sulfur either from inorganic sulfate or from organosulfur compounds such as sulfonates, sulfate esters, or sulfur-containing amino acids. Potential genes with similarity to those for the uptake of methylthioribose, arylsulfonate, sulfur ester, alkene sulfonate or taurine were not detected in the genome ofA. ferrooxidans. However, candidate genes potentially encoding uptake of inorganic sulfate were discovered, including a sulfate permease belonging the SulP family of the major facilitator superfamily of transporters (MFS) and a possible sulfate/thiosulfate/molybdenate ATP binding cassette (ABC)-type transporter.
The proposed SulP exhibits eleven predicted transmembrane helices and shows substantial conservation of amino acid sequence and predicted topology of transmembrane regions and extramembrane loops with similar proteins from a variety of neutrophilic bacteria. SulP contains the following motifs that have been identified for sulfate uptake: COG 0659, PFAM 00916 and eMOTIF IPB001902B. It also matched the sulfate uptake Tigerfam 00815, although the score was below the trusted cutoff.SulPis separated by ten base pairs from an upstream gene potentially encoding a carbonic anhydrase gene (cab1). The putative Cab1 ofA. ferrooxidansexhibits the Prosite signature (PS00704) of a beta-type, prokaryotic-type carbonic anhydrase. Carbonic anhydrase catalyzes the formation of HCO3from CO2+ H+. Whereas SulP functions as a H+symport in most organisms  the juxtaposition of the putativecab1andsulPinA. ferrooxidansin a potential transcriptional unit might facilitate the function of SulP as a HCO3-antiporter rather than as a H+symporter. Supporting this conjecture is experimental evidence for a SulP-like sulfate-HCO3-antiporter in rat . In addition, linkage of this gene pair is conserved in several organisms and, in the case ofLeptospira interrogans, the couple appear as a gene fusion (data not shown), suggesting that the association represents a functional coupling.
An alternative hypothesis is that SulP/Cab1 can function in the reverse direction, exporting sulfate in order to import carbonate for the enhancement of CO2fixation. Supporting this idea is the finding that carbonic anhydrase activity is associated with CO2fixation in a number of autotrophic organisms .
The suggested ABC-type membrane uptake mechanism for sulfate/thiosulfate/molybdate uptake inA. ferrooxidansconsists of a membrane spanning protein (CysU), that could form a channel for sulfate, and a membrane associated ATP-binding protein (CysA) that could provide energy for the transmembrane transport process. CysU is predicted to contain six transmembrane regions that align with similar regions in similar proteins. The genes cysUandcysApotentially form an operon-like structure (data not shown). Typically, sulfate is delivered to the ABC-uptake pump by a high affinity soluble periplasmic protein . Two candidate genes for this function were detected inA. ferrooxidans.One of them is embedded in a cluster of genes associated with molybdate uptake and is, therefore, more likely to be a molybdate transporter. Both candidates are members of a large paralogous gene family of ABC transporters  and the identification of their functions awaits experimental investigation. However, bioinformatics analysis has pinpointed them as the most likely candidates for sulfate transporters thus enabling a focus for the experimental scientist.
SECTION 3: Proposed pathway for sulfur assimilation into cysteine via APS
Activation of sulfur via APS
After entry into the cell, sulfate is generally made biologically available by activation with ATP in a reaction catalyzed by ATP sulfurylase, yielding APS and PPi. ATP sulfurylase is a heterodimeric enzyme encoded by genes termedcysDandcysNin many organisms . Potential orthologs ofcysDandcysNwere identified inA. ferrooxidans(Table 1). In addition, experimental evidence demonstrates that sulfate can be incorporated into APS inA. ferrooxidans.
Reduction of APS to sulfite
In plants, phototrophic bacteria and some other bacteria, sulfur is subsequently reduced to sulfite by APS reductase. However, in other organisms, including enteric bacteria, cyanobacteria and yeast, APS is further activated by another ATP to form PAPS that is, in turn, reduced to sulfite by PAPS reductase. APS and PAPS reductases show some 25–30% amino acid sequence identity, including a highly conserved (KRT)ECG(LS)H signature making their differentiation by bioinformatic analysis more difficult. However, it has been suggested that they can be distinguished by the presence of key motifs including two pairs of cysteine residues, characteristically found in APS reductase but not in PAPS reductase, that coordinate a 4Fe-4S center [16,17].A. ferrooxidanshas a potential gene (termedcysH) whose putative protein product not only exhibits the conserved (KRT)ECG(LS)H signature of the APS/PAPS reductase families but also the critical two cysteine pairs diagnostic of APS reductase (Figure2). Therefore, we tentatively identify it as an APS reductase.
A potential gene, termedplr(PAPS-like reductase or Cys-H-like), was found in the genome ofA. ferrooxidanswhose putative protein has weak similarity with PAPS reductase. It also exhibits weak similarity (e-11) to COG0175 (PAPS reductase) and weak similarity to Tigerfam00434 (PAPS reductase) but with a score below the trusted cutoff. An alignment of its hypothetical amino acid sequence with characterized APS and PAPS reductases reveals that it lacks the conserved signature (KRT)ECG(LS)H found in both APS and PAPS reducatses (Figure2). However, it does have two of the four critical cysteines displayed by APS reductase (highlighted in blue and bold, Figure2).Plris located in a gene cluster (which will be described in section 5.3) that containscysBand this association raises the possibility that the PAPS-like reductase may play a role in sulfur metabolism. However, at present its function remains unknown.
Reduction of sulfite to sulfide
After its formation, sulfite is typically reduced to sulfide by sulfite reductase. In bacteria, sulfite reductase is an oligomeric enzyme with a subunit composition of alpha(8)-beta(4) . The alpha subunit is a flavoprotein encoded bycysJwhile the beta component is a siroheme, iron-sulfur protein encoded bycysI. CandidatecysIandcysJwere identified inA. ferrooxidans.The putative CysI exhibits the prosite motif PS00365 [STV]-G-C-x(3)-C-x(6)-[DE]-[LIVMF]-[GAT]-LIVMF] characteristic of the beta subunit of sulfite reductase, in which the two cysteines are involved in the coordination of the iron sulfur center (Figure3A). CysI also contains Pfam motifs PF0340 and PF01077 characteristic of nitrite/sulfite reductase in which PF01077 contains the siroheme binding motif as well as the iron-sulfur cysteine ligands (Figure3A). The putative CysJ contains Pfam motifs characteristic of the alpha subunit of sulfite reductase including PF00258 (flavodoxin domain), PF00175 (NAD-binding domain) and PF00667 (FAD binding domain) [19,20] (Figure3B).
A possiblecysGwas also detected inA. ferrooxidans(Table 1 - seeadditional file 1) potentially encoding a uroporphyrin-III C-methyltransferase that has been shown in other organisms to be involved in siroheme formation and is, therefore, involved in the formation of CysJ. In addition to its bioinformatic identification, there is experimental evidence for the presence of a gene encoding this activity inA. ferrooxidans.
Formation of cysteine from O-acetyl-L-serine and sulfide
The final step of sulfur assimilation into cysteine is the synthesis of L-cysteine from O-acetyl-L-serine and sulfide catalyzed by O-acetyl-L-serine (thiol)-lyase encoded bycysMinE. coli. AcysMcandidate gene was identified inA. ferrooxidans(Table 1 - seeadditional file 1). In most organisms O-acetyl-L-serine is synthesized from serine by the activity of serine acetyltransferase encoded bycysE. Two potentialcysEgenes,cysE1andcysE2, were detected inA. ferrooxidansembedded in gene clusters predicted to be involved in the formation of Fe-S clusters and in nitrogen fixation. The possible significance of these genetic contexts with respect to the regulation of cysteine biosynthesis is discussed below.
Genetic organization and regulation of thecysoperon
The proposed genescysJIHDNGthat are speculated to be involved in the assimilation of sulfur into cysteine inA. ferrooxidanswere found to be arranged in an operon-like organization denominated thecysoperon. The proposecysoperon exhibits considerable conservation of gene content when compared to similar operons in other organisms (Figure4). Experimental evidence from several organisms demonstrates that the transcription of thecysoperon is under the control of the positive regulator CysB. CysB belongs to the LysR family of prokaryotic transcriptional regulators that share a highly conserved helix-turn-helix DNA recognition and binding domain . CysB is able to autorepress its own expression .
Regulatory binding sites for CysB upstream of thecysoperon have been identified inSalmonella typhimurium.A. ferrooxidanshas two candidate copies ofcysB(termedcysB1andcysB2, Table 1 [additional file 1]) that share 95% amino acid sequence identity. They encode potential proteins that exhibit significant conservation of amino acid sequence with characterized CysB proteins and also display the helix-turn-helix and LysR substrate binding motifs characteristic of the LysR family (Figure5). The genomic locations of the two hypotheticalcysBgenes will be discussed in section 5.3.
An additional facet ofcysoperon regulation is that transcriptional activation via CysB requires the presence of the coinducer N-acetylserine. InE. coliandS. typhimuriumit has been shown that a reduction in the availability of sulfide leads to the accumulation of O-acetyl-L-serine that spontaneously isomerizes to N-acetyl-L-serine. This latter compound binds to CysB inducing a conformational change that permits the protein to interact with its regulatory sites, upregulating thecysoperon and restoring sulfide levels [22,25,26]. O-acetyl-L-serine is synthesized via the acetylation of serine by the enzyme serine acetyltranferase that is the product ofcysE. It is speculated that the hypothetical CysE1and CysE2, previoiusly mentioned in the formation of cysteine, are also responsible for the formation of N-acetyl-L-serine and therefore play an important role in the regulation of thecysoperon.
A second mechanism regulatingcysoperon activity has been proposed . Excess cysteine is toxic because it promotes oxidative DNA damage by promoting the Fenton reaction  and the synthesis of cysteine can be down-regulated to prevent excess buildup via feedback inhibition on serine acetyltransferase (CysE) by cysteine. This, in turn, limits the production of O-acetyl-L-serine and, since this compound is one of the precursors of cysteine, its diminution causes a concomitant decrease in the synthesis of cysteine. A reduction of CysE activity is also predicted to reduce CysB activity that, in turn, will downregulate the expression of thecysoperon.
The notable conservation of the genes and their organization in the proposedcysoperon, taken together with the conservation of both the predicted CysB transcriptional activator and CysE co-activator, strongly suggests thatA. ferrooxidanscan regulate thecysoperon in a manner comparable to that described for other organisms. This speculation should provoke future bioinformatic and experimental searches for CysB regulatory binding sites in appropriate regions of theA. ferrooxidansgenome. Given this conservation and the fact that the pH of the cytoplasm ofA. ferrooxidansis near neutral suggests that the bioinformatic predictions may be open to experimental validation by complementation of well characterized mutants in other organisms such asE. coli(21, 29).
SECTION 4: Proposed sulfation of metabolites via the PAPS pathway
Some organisms, such as enteric bacteria and fungi phosphorylate APS with a second ATP to yield PAPS using the enzyme adenylylsulfate kinase (CysC). A discretecysCwas not found in the genome ofA. ferrooxidans. However, two non-identical copies of putative genes, termedcysNC1andcysNC2, were detected that appear to be fusions ofcysNandcysC.CysNC2is adjacent tocysD2that potentially encodes the small subunit of ATP sulfurylase (Figure6).CysD2represents a second, non-identical copy ofcysD, the other (cysD1) is associated with the proposedcysoperon described earlier.
The hypothetical CysNC1 and CysNC2 ofA. ferrooxidansexhibit several motifs and key conserved amino acids characteristic of the fused ATP sulfurylase/kinases of other organisms (Figure6), suggesting that they may carry out a role in PAPS biosynthesis similar to that determined in these organisms . It has also been demonstrated that a segment ofA. ferrooxidansDNA, prepared from a randomly cloned genomic library, complemented anE. colimutant deficient in CysC activity  providing experimental support for the bioinformatic identification of a possible gene encoding ATP sulfurylase/kinase activity.
FusedcysN/cysCgenes have been described inRhizobium sp.where the fusion is termednodQ. InR. meliloti, NodP (similar to theA. ferrooxidansCysD) and NodQ (similar to theA. ferrooxidansCysNC) are required for the synthesis of PAPS which provides activated sulfur for the modification of the reducing end of Nod, a factor essential for the formation of root nodules . In addition, PAPS is used inS. melilotifor the sulfation of cell surface polysaccharides . There are two copies ofnodPQinR. meliloti,which are functionally redundant in the sense that both copies have to be inactivated to impair nodule formation in alfalfa . PAPS also serves as a substrate in Mycobacteria for the biosynthesis of sulfolipids that may function as virulence factors . In general, sulfated molecules are directed outside the cell and act as modulators of cell-cell communication and cell adhesion and PAPS serves as a universal intermediate in their biosynthesis. Given the juxtaposition ofcysNC1with a gene potentially encoding a fibronectin-like binding protein that may be involved in cell adhesion and the juxtaposition ofcysNC2with a cluster of genes potentially encoding functions involved in capsular biosynthesis (Figure6), we speculate thatA. ferrooxidansalso utilizes PAPS as a donor for the sulfation of extra-cellular proteins which may be involved in adhesion of the organism to substrates such as sulfur and pyrite . Genes potentially encoding the SecA translocase system and the PapCD chaperone-usher system were detected in the genome ofA. ferrooxidans(data not shown), suggesting that the organism has the capacity to export and assemble adhesive structures.
An alternative, but not necessarily mutually exclusive, role for the function of the fused ATP sulfurylase/kinase protein has been suggested from studies of the kinetic constants of theA. aeolicusenzyme. This enzyme appears to be optimized, not for the formation of PAPS, but in the reverse direction generating ATP and sulfate. Consequently, in addition to its postulated role in the formation of PAPS, it may function in the opposite direction as a route to recycle PPi produced by biosynthetic reactions and for the formation of sulfate in the absence of an inorganic source of sulfate . Therefore, APS appears to lie at a critical metabolic branch point  channeling sulfur to cysteine or to PAPS, where PAPS can serves as a substrate for sulfation of other metabolites, or, in the reverse direction for recycling PPi and sulfate. It will be important in the future to investigate these predictions and to determine the regulatory mechanisms that control the flux of sulfur to the two pathways.
Consistent with the contention thatA. ferrooxidansis capable of PAPS formation is the presence of a candidatecysQ, associated with thecysD2NC2gene cluster, but separated from it by a gene potentially encoding a transposase (Figure6). CysQ encodes a 3'(2'),5'-bisphosphate nucleotidase that removes the 3'-phosphate from APS to regenerate AMP, helping to drive the APS-PAPS cycle and diminishing the toxicity of PAPS . We speculate that the hypothetical CysQ performs a similar function inA. ferrooxidans.
The candidate CysQ ofA. ferrooxidansexhibits the two prosite motifs (PDOC00547) characteristic of the enzyme as shown in the following two segments of the predicted amino acid sequence where the prosite motifs are underlined: WGFFWLVDPLDGTKEFIRKNGEYT and LGPTMEWDTAAAQIIATESGCRVEA the prosite patterns are: [FWV]-x(0,1)-[LIVM]-D-P-[LIVM]-D-[SG]-[ST]-x(2)-[FYA]-x-[HKRNSTY] where the first D and the T bind a metal ion and [WYV]-D-x-[AC]-[GSA]-[GSAPV]-x-[LIVFACP]-[LIVM]-[LIVAC]-x(3)-[GH]-[GA], (the final [GA] is replaced by a [C] in theA. ferrooxidanssequence).
SECTION 5: Formation of iron-sulfur centers and the proposed connection of sulfur metabolism with nitrogen fixation, hydrogen oxidation and sulfur reduction
Proposediscoperon for the formation of Fe-S centers
Cysteine serves as the main source of sulfur for the biosynthesis of Fe-S centers. Proteins with Fe-S centers are key catalysts in metabolism and are central for many processes that involve electron transfer. Fe-S proteins are also involved in sensing and transcriptional control . One route for Fe-S cluster biogenesis is exemplified by a protein complex encoded by the iscSUA operon involved in the formation nitrogenase and other proteins containing Fe-S centers . In this system,iscS(also known asnifS) encodes a cysteine desulfurase that removes an active sulfur from cysteine andicsUandiscAencode proteins that are thought to serve as a scaffold for Fe-S center assembly. Also, frequently forming part of the operon arehscAandhscBthat encode chaperones thought to aid in Fe-S center assembly, andfdxencoding a ferredoxin. The negative regulatoriscRis generally found upstream of the operon. IscR has been shown to contain a 2Fe-2S center that appears to be required for IscR to bind to its regulatory site and repress transcription. A model has been proposed in which IscR can sense the level of Fe-S center formation and adjust the expression of theiscoperon in accordance with requirements for Fe-S center formation .
A. ferrooxidanshas a putativeiscgene complex that we predict is an operon and which exhibits considerable conservation of organization compared to several organisms (Figure7). The proposed operon is also predicted to contain two genes involved in tRNA modification, one of which (trmH) is a predicted to encode a tRNA (Gm18) 2'-O-methyltransferase activity (Tigerfam00050)  and the other (trmU) potentially encodes a 5-methylaminomethyl-2-thiouridylate methyl transferase that transfers a methyl-thio group from SAM to tRNA (Tigerfam00420). As well as its role in Fe-S center formation, IscS has also been shown to be involved in the addition of thionucleotides to RNA  which might explain the association oftrmUwithiscS1. In addition, embedded in the proposediscoperon, is a hypothetical gene of unknown function andcysE1, one of the two copies ofcysEin the genome ofA. ferrooxidans. The possible significance of the location ofcysE1in theiscoperon will be discussed in section 6.
Additional copies ofiscSin a proposed nitrogen fixing regulon
A. ferrooxidansexhibits two additional copies of the cysteine desulfurase geneiscS(termedisc2and3) embedded in two adjacent but divergent gene clusters that include genes predicted to be involved in nitrogen fixation (Figure8). One of the clusters also containscysE2, the predicted second copy ofcysE, and additional copies of the Fe-S center biosynthesis genesiscU2andiscA2.The genecysEis also found embedded in thenifcluster inSynechococcus sp.and other nitrogen fixing bacteria .
CysB1, encoding the positive regulator of the cys operon, was found juxtaposed to a predictedhoxA(hydrogen uptake regulator) (Figure9).HoxAis a member of the AtoC two component response regulator family of transcriptional activators, containing CheY-like receiver, AAA-type ATPase and DNA binding domains (COG 2204.1).HoxAhas significant similarity to otherhoxAgenes involved in the regulation of the hox regulon (hydrogen uptake and oxidation) . HoxA has also been shown to be a negative regulator of its own expression . Adjacent, but divergently oriented tocysB1-hoxA, is a potential twenty-one gene cluster termed thehox1cluster, (Figure9A) predicted to be involved in hydrogen oxidation and hydrogenase formation, supporting the contention thathoxAregulates the expression of thehox1cluster. A second predicted cluster of hydrogenase expression and formation genes (hox2) that is not linked tohoxAwas also detected in the genome (Figure9A) but it is not known if this regulon could also be controlled byhoxA.
Hydrogen is an obligatory by-product of nitrogen fixation in a reaction that consumes about 27% of the electron flux. Some organisms export the hydrogen, while others have genes that encode hydrogen utilization functions permitting them to recover usable energy from the hydrogen, thereby increasing their metabolic efficiency . In addition, a number of organisms can oxidize hydrogen derived from the environment. Hydrogen oxidation can proceed aerobically in which electrons withdrawn from hydrogen reduce oxygen to water via a series of cytochromes resulting in the generation of a proton motif force for energy transduction. Hydrogen oxidation can also occur anaerobically in which the electrons from hydrogen are fed into the APS pathway via a protein complex encoded by thehmcoperon (high molecular weight cytochrome) resulting in the reduction of sulfur to sulfide . The HMC complex includes a protein with an iron-sulfur center of the "bacterial-type" 4Fe-4S ferredoxins. A predicted gene cluster potentially encoding an HMC complex was detected in theA. ferrooxidansgenome (data not shown) and we propose thatA. ferrooxidanshas the capacity to feed electrons from hydrogen oxidation into the sulfur reduction pathway.
SinceA. ferrooxidansfixes nitrogen  it presumably generates hydrogen in the process. Various strains ofA. ferrooxidanshave also been shown to oxidize externally supplied hydrogen both aerobically and anaerobically using, in the latter case, Fe+3or S0as electron acceptors [49,50] Unexpectedly, hydrogen did not support the growth of the strain ofA. ferrooxidansused in the genome sequencing studies (type strain ATCC23270)  despite the predicted presence of the required hydrogen utilization genes. One possible explanation is that the predicted hydrogen utilization genes are used only for hydrogen oxidation during nitrogen fixation and cannot support growth using an exogenous hydrogen source. Alternatively, the right physiological conditions might not have been used in the experimental design to provoke the activity of the genes. The strong bioinformatic predictions for the presence and conservation of two potential gene clusters predicted to be involved in hydrogen utilization suggest that further experimental work is warranted to reexplore the issue of hydrogen utilization in the type strain ofA. ferrooxidans.
The significance of the proposed transcriptional coupling between the predictedcysBandhoxAgenes requires investigation. We speculate that simultaneous transcriptional activation ofcysBandhoxApromotes the concomitant expression of the hox1 regulon (and possible also the hox2 regulon) and thecysoperon. Activation of thecysoperon will, in turn, provide additional enzymatic activity to absorb the electron flux delivered by the oxidation of hydrogen. It will also provide increased amounts of cysteine to serve as a source of sulfur for the formation of Fe-S centers used by proteins involved in nitrogen fixation and hydrogen oxidation. In other organisms CysB and HoxA are autorepressors and their predicted juxtaposition in theA. ferrooxidansgenome raises additional questions regarding their possible autoregulation and their role in global regulatory control of hydrogen utilization and sulfur metabolism.
A second copy of the cys operon regulator cysB (cysB2) is found adjacent to a gene termedtcrr(Fig.9B).Tcrrexhibits the CheY-like receiver and DNA binding domains of the AtoC (COG2204.1) family of two component response regulators but it lacks the AAA-like ATPase domain characteristic of this family. It has weak similarity to HoxA (expectancy in Blast = 10-8)We speculate that the co-transcription oftcrrandcysB2could simultaneously activate the unknown gene targets of thetcrrAtranscriptional regulator and also the cys regulon via the activity ofcysB2. Tcrr is predicted to be the receiver partner of a two component transcriptional regulatory system but the activator of this partnership is not known. A consideration that may shed light on the significance of the juxtaposition of thecysB2-tcrrcouple is that a potential gene,cobS, was detected that may lie within the same transcriptional unit ascysB2-tcrr(Figure9B). CobS is predicted to be involved in cobalmin biosynthesis which is required,inter alia, for the conversion of homocysteine to methionine. This impinges on sulfur metabolism because cysteine serves as the source of sulfur for methioine biosynthesis . Furthermore, downstream fromcobS, but separated from it by three open reading frames whose predicted proteins have no matches in GenBank, is the putativecysH-like geneplrthat was described earlier. Our bioinformatic analysis underscores the need to evaluate the expression and function ofplrand the predicted orphan genes since these may encode sulfur related functions peculiar toA. ferrooxidansand, therefore, merit special attention. The best hit ofplris to a phage protein (Table 1 - [seeadditional file 1]) suggesting that this region may have entered theA. ferrooxidansgenome by lateral gene transfer.
SECTION 6. Proposed integration of the regulation of sulfur assimilation with nitrogen fixation, hydrogen oxidation and sulfur reduction
Suggestions were presented above regarding possible mechanisms for the regulation of the assimilation of sulfur into cysteine in which a proposal was made connecting the activity ofcysB,cysEandiscSto nitrogen fixation and concomitant oxidation of hydrogen and reduction of sulfur. Our metabolic reconstruction also offers an initial and preliminary view of how these components might be integrated into a regulatory network providing a first approximation of the global control of sulfur metabolism inA. ferrooxidans(Figure10). This overview is extremely speculative, but we hope it might serve to highlight possible regulatory mechanisms worthy of experimental investigation.
At the heart of the proposed regulatory network is the pathway from sulfate to cysteine via APS, catalyzed by thecysoperoncysJIHDNunder the control of the positive transcriptional regulator CysB. In the event that the cell requires activation of the nitrogen fixing (nif) operon, the need for sulfur for the Fe-S centers of nitrogenase could be met by activation of theisccluster in whichiscS2andiscS3encoding cysteine desulfurase are embedded (Figure8). The withdrawal of sulfur by cysteine desulfurase will lead to the depletion of cysteine. However, activation of thenifandiscoperons is predicted to activate thecysEgenes embedded in these operons resulting in the production of N-acetyl-L-serine that, in turn, will activate thecysoperon via CysB helping to restore cysteine levels. Nitrogen fixation takes place anaerobically or microaerophilically, an important point that we will return to in the next paragraph.
The hydrogen byproduct of nitrogen fixation is predicted to activate one or both of the hydrogen regulons ofA. ferrooxidansviahoxA. Hydrogen utilization requires proteins with Fe-S centers calling again for products of theiscoperon. In addition, since the cell is under anaerobic conditions, the electrons from the oxidation of hydrogen will be fed into the APS pathway thereby reducing sulfate to some form of reduced sulfur, such as polysulfate, or polythionate which could be stored for later use as an energy and electron source when aerobic oxidation is restored. Electrons may also be used to reduce sulfate to sulfide. The concomitant increase of electron flux in the sulfur reduction pathway probably requires greater activity of thecysoperon genes to provide the additional enzymatic activity to absorb the flux.
Associated with hoxA, the proposed transcriptional regulator of the hox1, is one of the two copies ofcysB. Increased CysB activity, coupled with the proposed up-regulated activity of CysE, resulting from the aforementioned activation of thenifregulon, is predicted to increase levels ofcysoperon expression that, in turn, will help to cope with the increased electron flux from hydrogen oxidation. In addition, transcriptional activation oftcrrcould lead to increased expression of bothcobS, involved in cobalmin biosynthesis and thus methionine formation, and also of the putative geneplrwith weak similarity tocysH.
The presence of redundant copies of bothcysBandcysEcould reflect the need for alternative routes to activate thecysoperon depending on different regulatory signals. This possibility merits further investigation. The model of sulfur regulation also highlights other nodes of potential regulation that deserve further exploration. For example, what are the mechanisms that dictate the flow of activated sulfur to cysteine or to sulfated metabolites at the proposed APS decision point? What regulates the expression ofcysQthat is predicted to convert PAPS to APS? Absent from the model, at the present time, are the pathway connections and regulatory mechanisms that could incorporate sulfur oxidation, iron oxidation and iron homeostasis into the description of sulfur assimilation. However, their integration into the model awaits a more adequate description of these processes inA. ferrooxidans.
We posit that the preliminary metabolic model for sulfur uptake and assimilation inA. ferrooxidanspresented here has permitted the capture of potentially biologically valid relations that have now been highlighted for experimental validation. By deconvoluting aspects of the metabolic pathway and control mechanisms involved in sulfur metabolism and how these might be integrated with other metabolic processes such as nitrogen fixation, hydrogen utilization and sulfur reduction, we hope to have provided a glimpse of the autotrophic physiology that makesA. ferrooxidansof particular interest. It is also hoped that the work of the experimental scientist has been facilitated by this study, enabling a focus on the possible and likely, pinpointing areas of particular interest and reducing the labor in search of less likely pathways and mechanisms.
The proposed metabolic model provides an important preliminary step in understanding global sulfur metabolism inA. ferrooxidansespecially given the extreme difficulties involved in its genetic manipulation and biochemical analysis. It identifies some of the easier and more conspicuous aspects of the regulation of sulfur assimilation. Enfolded within the suggested model are much more complex and subtle mechanisms of regulation whose understanding will require additional bioinformatic analysis and experimental investigation.
Known metabolic pathways of sulfur assimilation were obtained from the MetaCyc database , KEGG  and ERGO . Amino acid sequences derived from genes identified as being involved in sulfur assimilation were used as query sequences to search the partial genome sequence ofA. ferrooxidansATCC 23270 in the TIGR (preliminary sequence data was obtained from The Institute for Genomic Research through the website athttp://www.tigr.org) and ERGO  data bases using TblastN and BlastP , respectively, with default parameters. When a prospective candidate gene was identified its predicted amino acid sequence was used to formulate a BlastP search of the non-redundant data base at NCBI. Only bidirectional best hits were accepted as evidence for putative orthologs. Candidate genes and their translated proteins were further characterized employing the following bioinformatic tools available in the web: primary structure similarity relations ClustalW , secondary structure predictions PSI-PRED  and SSpred , transmembrane predictions TMHMM , motif predictions Prosite  and PRINTS , domain predictions ProDom , Pfam  and Tigrfam , prediction of protein localization sites Psort . Information regarding the genetic organization of genes in organisms other thanA. ferrooxidanswere obtained from Integrated Genomics .
Work supported by Fondecyt No. 1010623. We thank the Institute of Genome Research (TIGR) and Integrated Genomics, Inc. (IG) for the use of their partial sequence of theA. ferrooxidansgenome. Sequencing ofA. ferrooxidanswas accomplished with support from the USA Dept. of Energy (DOE).
Laboratory of Bioinformatics and Genome Biology, University of Santiago (USACH)
Program of Cellular and Molecular Biology, I.C.B.M., Faculty of Medicine, University of Chile
Millennium Institute of Fundamental and Applied Biology
Rawlings DE:Heavy metal mining using microbes.Annu Rev Microbiol2002,56:65–91.View ArticlePubMed
Rohwerder T, Sand W:The sulfane sulfur of persulfides is the actual substrate of the sulfur-oxidizing enzymes from Acidithiobacillus and Acidiphilium spp.Microbiology2003,149:1699–1710.View ArticlePubMed
Ramirez P, Toledo H, Guiliani N, Jerez CA:An exported rhodanese-like protein is induced during growth of Acidithiobacillus ferrooxidans in metal sulfides and different sulfur compounds.Appl Environ Microbiol2002,68:1837–1845.View ArticlePubMed
Masau RJ, Oh JK, Suzuki I:Mechanism of oxidation of inorganic sulfur compounds by thiosulfate-grown Thiobacillus thiooxidans.Can J Microbiol2001,47:348–358.View ArticlePubMed
Sugio T, Hirose T, Ye LZ, Tano T:Purification and some properties of sulfite:ferric ion oxidoreductase from Thiobacillus ferrooxidans.J Bacteriol1992,174:4189–4192.PubMed
Tuovinen OH, Kelley BC, Nicholas DJ:The uptake and assimilation of sulphate by Thiobacillus ferrooxidans.Arch Microbiol1975,105:123–127.View ArticlePubMed
Kusano T, Sugawara K, Inoue C, Takeshima T, Numata M, Shiratori T:Electrotransformation of Thiobacillus ferrooxidans with plasmids containing a mer determinant.J Bacteriol1992,174:6617–6623.PubMed
Liu Z, Guiliani N, Appia-Ayme C, Borne F, Ratouchniak J, Bonnefoy V:Construction and characterization of a recA mutant of Thiobacillus ferrooxidans by marker exchange mutagenesis.J Bacteriol2000,182:2269–2276.View ArticlePubMed
Selkov E, Overbeek R, Kogan Y, Chu L, Vonstein V, Holmes D, Silver S, Haselkorn R, Fonstein M:Functional analysis of gapped microbial genomes: amino acid metabolism of Thiobacillus ferrooxidans.Proc Natl Acad Sci USA2000,97:3509–3514.View ArticlePubMed
Barreto M, Quatrini R, Bueno S, Arraigada S, Valdes J, Silver S, Jedlicki E, Holmes DS:Aspects of the predicted physiology of Acidithiobacillus ferrooxidans deduced from an analysis of its partial genome sequence.Hydrometallurgy2003,71:97–105.View Article
Tripp BC, Smith K, Ferry JG:Carbonic anhydrase: new insights for an ancient enzyme.J Biol Chem2001,52:48615–48618.View Article
Bissig M, Hagenbuch B, Stieger B, Koller T, Meier PJ:Functional expression cloning of the canalicular sulfate transport system of rat hepatocytes.J Biol Chem1994,269:3017–3021.PubMed
Jacobson BL, He JJ, Vermersch PS, Lemon DD, Quiocho FA:Engineered interdomain disulfide in the periplasmic receptor for sulfate transport reduces flexibility. Site-directed mutagenesis and ligand-binding studies.J Biol Chem1991,266:5220–5225.PubMed
Holland B, Blight MA:ABC-ATPases, adaptable energy generators fuelling transmembrane movement of a variety of molecules in organisms from bacteria to humans.J Mol Biol1999,293:381–399.View ArticlePubMed
Leyh TS, Taylor JC, Markham GD:The sulfate activation locus of Escherichia coli K12: cloning, genetic, and enzymatic characterization.J Biol Chem1988,263:2409–2416.PubMed
Berendt U, Haverkamp T, Prior A, Schwenn JD:Reaction mechanism of thioredoxin: 3'-phospho-adenylylsulfate reductase investigated by site-directed mutagenesis.Eur J Biochem1995,233:347–356.View ArticlePubMed
Kopriva S, Buchert T, Fritz G, Suter M, Benda R, Schunemann V, Koprivova A, Schurmann P, Trautwein AX, Kroneck PM, Brunold C:The presence of an iron-sulfur cluster in adenosine 5'-phosphosulfate reductase separates organisms utilizing adenosine 5'-phosphosulfate and phosphoadenosine 5'-phosphosulfate for sulfate assimilation.J Biol Chem2002,277:21786–21791.View ArticlePubMed
Kredich NM:Biosynthesis of Cysteine.In: Escherichia coli and Salmonella2 Edition(Edited by: Neidhardt FC).ASM Press, Washington, DC 1996,1:514–527.
Nakayama M, Akashi T, Hase T:Plant sulfite reductase: molecular structure, catalytic function and interaction with ferredoxin.J Inorg Biochem2000,82:27–32.View ArticlePubMed
Crane BR, Siegel LM, Getzoff ED:Sulfite reductase structure at 1.6 Å: evolution and catalysis for reduction of inorganic anions.Science1995,270:59–67.View ArticlePubMed
Sugio T, Suzuki H, Tanaka T, Matsugi S, Tanaka K, Kanao T, Tano T:A DNA region that complements on Escherichia coli cysG mutation in Thiobacillus ferrooxidans.Biosci Biotechnol Biochem1995,59:728–730.View ArticlePubMed
Lochowska A, Iwanicka-Nowicka R, Plochocka D, Hryniewicz MM:Functional dissection of the LysR-type CysB transcriptional regulator. Regions important for DNA binding, inducer response, oligomerization, and positive control.J Biol Chem2001,276:2098–2107.View ArticlePubMed
Ostrowski J, Kredich NM:Negative autoregulation of cysB in Salmonella typhimurium: in vitro interactions of CysB protein with the cysB promoter.J Bacteriol1991,173:2212–2218.PubMed
Hryniewicz MM, Kredich NM:Hydroxyl radical footprints and half-site arrangements of binding sites for the CysB transcriptional activator of Salmonella typhimurium.J Bacteriol1995,177:2343–2353.PubMed
Hryniewicz MM, Kredich NM:The cysP promoter of Salmonella typhimurium : characterization of two binding sites for CysB protein, studies of in vivo transcription initiation, and demonstration of the anti-inducer effects of thiosulfate.J Bacteriol1991,173:5876–5886.PubMed
Monroe RS, Ostrowski J, Hryniewicz MM, Kredich NM:In vitro interactions of CysB protein with the cysK and cysJIH promoter regions of Salmonella typhimurium.J Bacteriol1990,172:6919–6929.PubMed
Park S, Imlay JA:High levels of intracellular cysteine promote oxidative DNA damage by driving the fenton reaction.J Bacteriol2003,185:1942–1950.View ArticlePubMed
Schwedock JS, Liu C, Leyh TS, Long SR:Rhizobium meliloti NodP and NodQ form a multifunctional sulfate-activating complex requiring GTP for activity.J Bacteriol1994,176:7055–7064.PubMed
Fry IJ, Garcia E:Cloning and caracterization of Thiobacillus ferrooxidans genes involved in sulfur assimilation.In Biohydrometallurgy(Edited by: Salley J, McCready RGL, Wichlacz PL).1989, 171–185.
Laeremans T, Coolsaet N, Verreth C, Snoeck C, Hellings N, Vanderleyden J, Martinez-Romero E:Functional redundancy of genes for sulphate activation enzymes in Rhizobium sp. BR816.Microbiology1997,143:3933–3942.View ArticlePubMed
Abola AP, Willits MG, Wang RC, Long SR:Reduction of adenosine–5'-phosphosulfate instead of 3'-phosphoadenosine–5'-phosphosulfate in cysteine biosynthesis by Rhizobium meliloti and other members of the family Rhizobiaceae.J Bacteriol1999,181:5280–5287.PubMed
Keating DH, Willits MG, Long SR:A Sinorhizobium meliloti lipopolysaccharide mutant altered in cell surface sulfation.J Bacteriol2002,184:6681–6689.View ArticlePubMed
Schwedock JS, Long SR:Rhizobium meliloti genes involved in sulfate activation: the two copies of nodPQ and a new locus, saa.Genetics1992,132:899–909.PubMed
Williams SJ, Senaratne RH, Mougous JD, Riley LW, Bertozzi CR:5'-adenosinephosphosulfate lies at a metabolic branch point in mycobacteria.J Biol Chem2002,277:32606–3215.View ArticlePubMed
Gehrke T, Hallmann R, Kinzler K, Sand W:The EPS of Acidithiobacillus ferrooxidans a model for structure-function relationships of attached bacteria and their physiology.Water Sci Technol2001,43:159–167.PubMed
Hanna E, MacRae IJ, Medina DC, Fisher AJ, Segel IH:ATP sulfurylase from the hyperthermophilic chemolithotroph Aquifex aeolicus.Arch Biochem Biophys2002,406:275–288.View ArticlePubMed
Neuwald AF, Krishnan BR, Brikun I, Kulakauskas S, Suziedelis K, Tomcsanyi T, Leyh TS, Berg DE:cysQ, a gene needed for cysteine synthesis in Escherichia coli K–12 only during aerobic growth.J Bacteriol1992,174:415–425.PubMed
Kiley PJ, Beinert H:The role of Fe-S proteins in sensing and regulation in bacteria.Current Opinion Microbiol2003,6:181–185.View Article
Yuvaniyama P, Agar JN, Cash VL, Johnson MK, Dean DR:NifS-directed assembly of a transient [2Fe–2S] cluster within the NifU protein.Proc Natl Acad Sci2000,97 :599–604.View ArticlePubMed
Frazzon J, Dean DR:Feedback regulation of iron-sulfur cluster biosynthesis.Proc Nal Acad Sci2001,98:14751–14753.View Article
Persson BC, Jager G, Gustafsson C:The spoU gene of Escherichia coli, the fourth gene of the spoT operon, is essential for tRNA (Gm18) 2'-O-methyltransferase activity.Nucleic Acids Res1997,25:4093–4097.View ArticlePubMed
Iciek M, Wodek L:Biosynthesis and biological properties of compounds containing highly reactive, reduced sulfane sulfur.Pol J Pharmacol2001,53:215–225.PubMed
Huang TC, Lin RF, Chu MK, Chen HM:Organization and expression of nitrogen-fixation genes in the aerobic nitrogen-fixing unicellular cyanobacterium Synechococcus sp. strain RF–1.Microbiology1999,145:743–753.View ArticlePubMed
Van Soom C, Verreth C, MJ Vanderleyden J:Identification of a potential transcriptional regulator of hydrogenase activity in free-living Bradyrhizobium japonicum strains.Mol Gen Genet1993,239:235–240.PubMed
Durmowicz MC, Maier RJ:Roles of HoxX and HoxA inbiosynthesis of hydrogenase in Bradyrhizobium japonicum.J Bacteriol179:3676–3682.
Lenz O, Bernhard M, Buhrke T, Schwartz E, Friedrich B:The hydrogen-sensing apparatus in Ralstonia eutropha.J Mol Microbiol Biotechnol2002,4:255–262.PubMed
Rossi M, Pollock WB, Reij MW, Keon RG, Fu R, Voordouw G:The HMC operon of Desulfovibrio vulgaris subsp. vulgaris Hildenborough encodes a potential transmembrane redox protein complex.J Bacteriol1993,175:4699–4711.PubMed
Mackintosh ME:Nitrogen fixation in Thiobacillus ferrooxidans.J Gen Microbiol1978,105:215–218.
Drobner E, Huber H, Stetter KO:Thiobacillus ferrooxidans, a facultative hydrogen oxidizer.Appl Environ Microbiol,1990,56:2922–2923.
Ohmura N, Sasaki K, Matsumoto N, Saiki H:Anaerobic respiration using Fe(3+), S(0), and H(2) in the chemolithoautotrophic bacterium Acidithiobacillus ferrooxidans.J Bacteriol2002,184:2081–2087.View ArticlePubMed
Roth JR, Lawrence JG, Bobik TA:Cobalmin (Coenzyme B12): Synthesis and Biological Significance.Annu Rev Microbiol1996,50:137–181.View ArticlePubMed
Kanehisa M, Goto S, Kawashima S, Nakaya A:The KEGG databases at GenomeNet.Nucleic Acids Res2002,30:42–46.View ArticlePubMed
Overbeek R, Larsen N, Walunas T, D'Souza M, Pusch G, Selkov EJr, Liolios K, Joukov V, Kaznadzey D, Anderson I, Bhattacharyya A, Burd H, Gardner W, Hanke P, Kapatral V, Mikhailova N, Vasieva O, Osterman A, Vonstein V, Fonstein M, Ivanova N, Kyrpides N:The ERGO genome analysis and discovery system.Nucleic Acids Res2003,31:164–171.View ArticlePubMed
Altschul SF, Madden TL, Scha!er AA, Zhang J, Zheng Z, Miller W, Lipman DJ:Gapped BLAST and PSIBLAST - A new generation of protein database search programs.Nucleic Acids Res1997,25:3389–3402.View ArticlePubMed
Higgins D, Thompson J, Gibson T, Thompson JD, Higgins DG, Gibson TJ:CLUSTAL W: improving the sensitivity of progressive multiple sequence alignment through sequence weighting, position-specific gap penalties and weight matrix choice.Nucleic Acids Res1994,22:4673–4680.View ArticlePubMed
Jones DT:Protein secondary structure prediction based on position-specific scoring matrices.J Mol Biol1999,292:195–202.View ArticlePubMed
Pollastri G, Przybylski D, Rost B, Baldi P:Improving the prediction of protein secondary structure in three and eight classes using recurrent neural networks and profiles.Proteins2002,47:228–235.View ArticlePubMed
Krogh A, Larsson B, vonHeijne G, Sonnhammer ELL:Predicting transmembrane protein topology with a hidden Markov model: Application to complete genomes.J Mol Biol2001,305:567–580.View ArticlePubMed
Hofmann K, Bucher P, Falquet L, Bairoch A:The PROSITE database, its status in 1999.Nucleic Acids Res1999,27:215–219.View ArticlePubMed
Attwood TK, Beck ME, Bleasby AJ, Parry-Smith DJ:PRINTS - A database of protein motif fingerprints.Nucleic Acids Res1994,22:3590–3596.PubMed
Servant F, Bru C, Carrère S, Courcelle E, Gouzy J, Peyruc D, Kahn D:ProDom: Automated clustering of homologous domains.Briefings in Bioinformatics2002,3:246–251.View ArticlePubMed
Bateman A, Birney E, Cerruti L, Durbin R, Etwiller L, Eddy SR, Griffiths-Jones S, Howe KL, Marshall M, Sonnhammer EL:The Pfam protein families database.Nucleic Acids Res2002,30:276–280.View ArticlePubMed
Haft DH, Selengut JD, White O:The TIGRFAMs database of protein families.Nucleic Acids Res2003,31:371–373.View ArticlePubMed
Nakai K, Horton P:PSORT: a program for detecting sorting signals in proteins and predicting their subcellular localization.Trends Biochem Sci1999,24:34–36.View ArticlePubMed
This article is published under license to BioMed Central Ltd. This is an Open Access article: verbatim copying and redistribution of this article are permitted in all media for any purpose, provided this notice is preserved along with the article's original URL.